Western European lineage
Forty-three SNPs (22 intronic, 21 exonic) were found in Western European populations, of which 5 are non-synonymous and 16 synonymous (Fig. 4 and Table S6). To assess whether variation in the Tshr gene can be explained by local seasons, SNP frequencies were correlated to different environmental proxies. The non-synonymous SNPs were not associated with the tested environmental proxies (Table S6). In the Western European lineage, 3/43 SNPs (exonic, synonymous) significantly correlated with pairwise geographical distance (Fig. 4A, C, Table S6), indicating that there is a lack of regional equilibrium(Hutchison & Templeton, 1999), and that an alternative approach may be used to detect selection, classifying SNPs that show clinal variation(Endler, 1977). SNP frequency weakly correlated to pairwise latitudinal, longitudinal and altitudinal difference for the majority of the observed SNPs (latitude: 2 exonic SNPs, longitude: 11 intronic and 5 exonic SNPs, altitude: 3 intronic and 3 exonic SNP) (Fig. 4D-L, Table S6). These findings show that geographical distance, latitude and altitude by themselves are bad predictors for genetic variation in the TSHR gene, despite the fact that annual photoperiod-food abundance patterns depend on all these parameters. Therefore, for each sample location, the pCPP at which grass growth is initiated in spring (at 5-10°C ambient temperature; Cooper, 1964; Peacock, 1975; Peacock, 1976) was deduced from local annual photoperiod-ambient temperature ellipsoids. pCPP at 6.6°C achieved highest number of significant SNPs. Therefore, the temperature threshold for grass growth initiation in spring was set at 6.6°C, and was used to deduce corresponding pCPP, which were calculated to vary between 10.19 and 15.40 hours of light /24 hours (Fig. 1 and Table S1). In Western European samples, 5 intronic and 7 exonic SNPs strongly correlated to pairwise difference in pCPP (Fig. 4M, Table S6). FSTvalues for these specific SNPs were high (ranging from FST = 0.032 to 0.310, mean: 0.166). All these significant mutations were, however, synonymous SNPs. Strongest associations with pCPP were found for intronic SNP-158 (G>C), -128 (T>C), and exonic SNP126 (A>G) (Fig. 4M-O, Table S6). It is expected that between Orkney Island and between mainland, some of the variation reflects isolation and genetic drift. Therefore, the same analysis was performed excluding the Orkney Island populations and revealed similar results. Pairwise multilocus FST-values were high for populations that differ highly in pCPP, while FST-values were low for populations with similar pCPP (Fig. 3).